Research on Multi-target Detection Method of Remote Sensing Images Based on Improved Yolov5
A remote sensing image target detection method based on improved Yolov5 is proposed. An efficient channel attention mech-anism is added in the backbone network to improve the learning efficiency of positive sample features, and a detection output feature map is added in the multi-scale feature fusion layer to improve the accuracy of detection. For the detection accuracy of small targets, the cross-entropy loss function with gradient modulus is used to train the model to alleviate the problem of positive and negative sample imbalance. The experimental results show that the improved method proposed in this paper achieves high detection accuracy and good generalization ability on the test data set, which is significantly improved as compared with the original Yolov5, and can also reach the level of real-time detection in the test environment. The proposed improved method has great application value in the fields of resource survey and emergency rescue.
remote sensing imagesYolov5attention mechanismloss function
万方数据
维普
